Qwen1.0，比较拉，最新的1.5在下面
下载源码
git clone https://gitee.com/qzl66/Qwen.git下载模型
下载至Qwen/Qwen/下
git clone https://www.modelscope.cn/qwen/Qwen-14B-Chat.git
git clone https://www.modelscope.cn/qwen/Qwen-7B-Chat.git创建环境
conda create -n qwen
conda activate qwen
# 清华源速度就是杠杠的
pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple修改代码
修改openai_api.py文件
# 530行左右修改默认端口、ip（建议0.0.0.0）、模型位置
default='Qwen/Qwen-7B-Chat',
default=8000,
default='0.0.0.0',
# 487行有个流式传输的bug
delay_token_num = max([len(x) for x in stop_words])
# 改为
if stop_words:
delay_token_num = max([len(x) for x in stop_words])
else:
delay_token_num = 0
# 如果多显卡修改：
model = AutoModelForCausalLM.from_pretrained(
args.checkpoint_path,
device_map=device_map,
trust_remote_code=True,
resume_download=True,
).eval()
# 改为（num_gpus=显卡数量）：
from utils import load_model_on_gpus
model = load_model_on_gpus(args.checkpoint_path, num_gpus=4)最后直接运行
python openai_api.pyQwen1.5
下载模型
git clone https://www.modelscope.cn/qwen/Qwen1.5-7B.git创建环境
pip install vllm
conda create -n Qwen1.5 python=3.9 -y
conda activate Qwen1.5
# Install vLLM with CUDA 12.1.
pip install vllm# 最后的pip包
Package Version
----------------------------- ------------
accelerate 0.28.0
addict 2.4.0
aiohttp 3.9.3
aiosignal 1.3.1
aliyun-python-sdk-core 2.15.0
aliyun-python-sdk-kms 2.16.2
annotated-types 0.6.0
anyio 4.3.0
appdirs 1.4.4
APScheduler 3.9.1
argos-translate-files 1.1.4
argostranslate 1.9.0
async-timeout 4.0.3
attrs 23.2.0
Babel 2.14.0
beautifulsoup4 4.9.3
cachelib 0.12.0
certifi 2024.2.2
cffi 1.16.0
charset-normalizer 2.1.1
click 8.1.7
cloudpickle 3.0.0
commonmark 0.9.1
crcmod 1.7
cryptography 42.0.5
ctranslate2 3.20.0
cupy-cuda12x 12.1.0
datasets 2.18.0
Deprecated 1.2.14
dill 0.3.8
diskcache 5.6.3
einops 0.7.0
exceptiongroup 1.2.0
expiringdict 1.2.2
fastapi 0.110.0
fastrlock 0.8.2
filelock 3.13.1
Flask 2.2.2
flask-babel 3.1.0
Flask-Limiter 2.6.3
Flask-Session 0.4.0
flask-swagger 0.2.14
flask-swagger-ui 4.11.1
frozenlist 1.4.1
fsspec 2024.2.0
gast 0.5.4
h11 0.14.0
httptools 0.6.1
huggingface-hub 0.21.4
idna 3.6
importlib_metadata 7.0.2
importlib_resources 6.3.1
interegular 0.3.3
itsdangerous 2.1.2
Jinja2 3.1.3
jmespath 0.10.0
joblib 1.3.2
jsonschema 4.21.1
jsonschema-specifications 2023.12.1
lark 1.1.9
libretranslate 1.3.13
limits 3.10.1
llvmlite 0.42.0
LTpycld2 0.42
lxml 5.1.0
MarkupSafe 2.1.5
modelscope 1.13.1
Morfessor 2.0.6
mpmath 1.3.0
msgpack 1.0.8
multidict 6.0.5
multiprocess 0.70.16
nest-asyncio 1.6.0
networkx 3.2.1
ninja 1.11.1.1
nltk 3.8.1
numba 0.59.1
numpy 1.26.4
nvidia-cublas-cu12 12.1.3.1
nvidia-cuda-cupti-cu12 12.1.105
nvidia-cuda-nvrtc-cu12 12.1.105
nvidia-cuda-runtime-cu12 12.1.105
nvidia-cudnn-cu12 8.9.2.26
nvidia-cufft-cu12 11.0.2.54
nvidia-curand-cu12 10.3.2.106
nvidia-cusolver-cu12 11.4.5.107
nvidia-cusparse-cu12 12.1.0.106
nvidia-nccl-cu12 2.18.1
nvidia-nvjitlink-cu12 12.4.99
nvidia-nvtx-cu12 12.1.105
oss2 2.18.4
outlines 0.0.36
packaging 23.1
pandas 2.2.1
pillow 10.2.0
pip 23.3.1
platformdirs 4.2.0
polib 1.1.1
prometheus_client 0.20.0
protobuf 5.26.0
psutil 5.9.8
pyarrow 15.0.2
pyarrow-hotfix 0.6
pycparser 2.21
pycryptodome 3.20.0
pydantic 2.6.4
pydantic_core 2.16.3
Pygments 2.17.2
pynvml 11.5.0
python-dateutil 2.9.0.post0
python-dotenv 1.0.1
pytz 2024.1
PyYAML 6.0.1
ray 2.10.0
redis 4.3.4
referencing 0.34.0
regex 2023.12.25
requests 2.28.1
rich 12.6.0
rpds-py 0.18.0
sacremoses 0.0.53
safetensors 0.4.2
scipy 1.12.0
sentencepiece 0.1.99
setuptools 68.2.2
simplejson 3.19.2
six 1.16.0
sniffio 1.3.1
sortedcontainers 2.4.0
soupsieve 2.5
sse-starlette 2.0.0
stanza 1.1.1
starlette 0.36.3
sympy 1.12
tiktoken 0.6.0
tokenizers 0.15.2
tomli 2.0.1
torch 2.1.2+cu121
torchaudio 2.1.2+cu121
torchvision 0.16.2+cu121
tqdm 4.66.2
transformers 4.39.0
transformers-stream-generator 0.0.4
translatehtml 1.5.2
triton 2.1.0
typing_extensions 4.10.0
tzdata 2024.1
tzlocal 5.2
urllib3 1.26.18
uvicorn 0.28.1
uvloop 0.19.0
vllm 0.3.3
waitress 2.1.2
watchfiles 0.21.0
websockets 12.0
Werkzeug 2.2.2
wheel 0.41.2
wrapt 1.16.0
xformers 0.0.23.post1
xxhash 3.4.1
yapf 0.40.2
yarl 1.9.4
zipp 3.18.1修改文件夹名称为qwen1.5-14b-chat，方便对接api
运行模型
# sh
cd /datas/ptyhon_app/Qwen1.5/Qwen
/home/.conda/envs/qwen1_5/bin/python -m vllm.entrypoints.openai.api_server --model qwen1.5-14b-chat --port 8000 --dtype=half --tensor-parallel-size 4 --gpu-memory-utilization 0.9 --max-model-len 4096好用，最后PO张图
