Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Baichuan13B vllm 效果很差 #202

Open
moseshu opened this issue Jan 4, 2024 · 1 comment
Open

Baichuan13B vllm 效果很差 #202

moseshu opened this issue Jan 4, 2024 · 1 comment

Comments

@moseshu
Copy link

moseshu commented Jan 4, 2024

启动脚本

vllm=v0.2.6
python -m vllm.entrypoints.openai.api_server --model=/opt/apps/models/baichuan-13B--trust-remote-code --gpu-memory-utilization 0.9 --tensor-parallel-size 2 

client

for chunk in client.chat.completions.create(model="llama",
    messages=[
        {"role": "user", "content": "我是一个科学家"},
    ],stream=True,
    max_tokens=40,):
    
    print(chunk.choices[0].delta.content, end="", flush=True)


输出结果如下:
None<|im_end| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |None

 这是数据组成的格式不对么?
@codernew007
Copy link

需要根据api_server.py实现,在request中增加top_p、repetable_penality等参数,另外需要设置chat-template为baichuan

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants